Meta Muse Puts AI Agents at the Center of a New Digital Distribution Battle

Meta’s Muse is turning the emerging AI-agent market into a much bigger question for the internet. Who controls the customer relationship when software, rather than a person, decides where to shop, what to buy, and which digital services to use?

Meta launched Muse on September 8 as a personal AI agent designed to perform tasks, rather than answer questions. It can browse the web, interact with services, handle multi-step tasks, and act on a user’s behalf, including for shopping and payments. According to Meta, Muse runs inside a dedicated secure virtual machine with its own browser.

The early commercial response has already split the market. Shopify is opening its commerce infrastructure to Muse, while Amazon has blocked the agent from shopping on Amazon.com. Meta has also expanded Muse’s network of commerce and payment connections, adding companies including Walmart, Best Buy, Sephora, and Wayfair, alongside Shop Pay and PayPal.

This dispute points to a potential shift in digital distribution. For years, websites, search engines, social platforms, and app stores controlled how businesses reached consumers. AI agents could become another major intermediary, but this time, the intermediary may act directly for the customer.

Moving beyond the AI assistant

Traditional AI assistants mainly respond to questions. Muse, by contrast, is designed to take action. Meta says users can ask Muse to handle tasks such as sending emails, booking travel, and completing other activities across the services they use. The agent operates through its own secure virtual machine and browser, while users control how much access it receives.

Shopping matters particularly here, since it turns the AI agent into a potential commercial gateway. At launch, Meta connected Muse to Stripe’s Link wallet. Stripe said US consumers can connect their Link accounts to Muse and allow the agent to make purchases. At more than one million businesses that accept Link, Muse can use the consumer’s saved payment method. For other businesses, Link can issue a single-use virtual card limited to the approved purchase.

That gives Muse more than product-discovery capabilities. It gives the agent a full path from search and recommendation to transaction.

Shopify opens its commerce infrastructure to Muse.

Shopify has taken a different approach from Amazon. Shopify’s infrastructure already supports agentic commerce, letting merchants make products available across AI channels. Its Agentic Storefronts system lets businesses manage how products appear in AI-driven shopping experiences.

Shopify’s current documentation says customers can complete purchases through Shopify-powered direct checkout on Meta surfaces, including Muse when available. Shop Pay can also appear as a wallet or checkout option within Muse. The company also gives merchants controls over whether direct checkout is enabled and how their products get distributed through its AI channels.

This model makes AI agents another sales channel, rather than an unwanted intermediary. For merchants, the attraction is straightforward. If consumers increasingly ask AI agents what to buy, businesses need their products to be discoverable wherever those conversations happen.

Amazon takes the opposite approach.

Amazon has blocked Meta’s Muse from shopping on Amazon.com. Users attempting to access Amazon through Muse have encountered a message stating that continued access by an unauthorized AI agent violates Amazon’s Conditions of Use. Amazon said it hadn’t authorized Muse to access its retail platform and had asked Meta to remove Amazon from the experience.

Amazon has raised concerns about how Muse interacts with its platform, including the fact that the agent doesn’t identify itself when browsing. Amazon has also pointed to privacy and security concerns involving customer credentials. Meta, for its part, has said Muse doesn’t have visibility into users’ passwords or payment methods and that credentials users supply are stored securely.

This disagreement isn’t simply about whether consumers can shop through an AI assistant. It’s about who gets to control the digital interface between a customer and a business. Amazon operates both a massive retail marketplace and a major advertising business. If an AI agent becomes the primary interface for product discovery, the traditional retailer-controlled shopping journey could change substantially.

Payments are becoming part of the AI battle.

Payments are emerging as a critical piece of agentic commerce. Muse launched with Stripe’s Link wallet, while Meta later announced support for Shop Pay and PayPal. Meta also expanded its network of shopping connectors to include Walmart, Best Buy, American Eagle Outfitters, DICK’S Sporting Goods, Fanatics, Gap, Michael Kors, Sephora, Ulta, and Wayfair.

The significance here runs deeper than simply adding more payment buttons. An AI agent needs several pieces of infrastructure to complete a transaction: product information, inventory and pricing, merchant access, checkout, payment authorization, identity and security, consumer approval, and returns and dispute handling. Payment providers therefore have a potentially important role in determining whether agentic commerce can operate at scale.

PayPal’s own research illustrates the issue well. Its 2026 Agentic Commerce Pulse Report found that data security was the top barrier to investment in agentic commerce among the large enterprises surveyed, while 95% expected payment providers to play a leading or supporting role in the sector.

AI agents could change digital distribution.

The bigger question raised by Muse isn’t simply whether people will use AI to shop. It’s whether AI agents become the interface through which consumers reach businesses in the first place.

Search engines already influence which websites consumers discover. Social platforms determine which content reaches users. App stores control access to mobile applications. AI agents could introduce another layer that makes recommendations, compares products, initiates transactions, and potentially manages the relationship after a purchase.

That creates a new dependency for businesses. A retailer could have a strong website, a large product catalogue, and established payment infrastructure, yet still struggle to reach customers if an AI agent doesn’t surface its products. At the same time, businesses may want control over how their brands, pricing, product information, and customer relationships appear inside AI systems.

Shopify has explicitly positioned its agentic commerce tools around product discovery and brand control, while Amazon’s decision to block Muse shows that some platforms may prefer to retain control over their own customer experience instead.

Meta is expanding Muse into a broader platform.

Meta isn’t treating Muse as a standalone shopping assistant. At Meta Connect 2026, the company announced additional connectors for Muse and said it would bring the agent to its AI glasses in the coming months. Meta also added services for shopping, travel, food delivery, and workplace productivity.

According to Meta, Muse will eventually work with services including Expedia and Instacart, while Notion, Granola, GitHub, and Box will support work-related tasks. This expansion matters because a more capable agent has more opportunities to become the starting point for everyday digital activity. Shopping could end up being only one part of that relationship.

Security and trust remain major constraints.

The more authority an AI agent receives, the greater the consequences when something goes wrong. An agent that only answers a question can provide incorrect information. An agent that can send an email, access an account, book a service, or purchase a product can create a real-world transaction instead.

Meta says it built Muse with security and privacy controls in mind, including its secure virtual machine and mechanisms that keep users in control of access. Meta also uses a separate monitoring system to oversee actions the agent takes.

That said, the broader industry is still working through questions around agent identity, permissions, authentication, data access, liability, and consumer consent. These issues will become increasingly important as AI agents move from recommendation engines to genuine transaction engines.

The battle is moving from AI models to access

The Muse-Amazon dispute illustrates a different phase of the AI competition. The first phase focused heavily on model capability: which company had the strongest model, the best reasoning, and the most useful assistant.

The next phase could be about access instead. Which businesses allow an AI agent to interact with their systems? Which payment networks support agent-initiated transactions? Who controls customer data? Who owns the relationship when the agent makes the recommendation? And who bears responsibility when an agent makes a mistake?

Meta is building Muse around a growing network of connectors. Shopify is extending commerce infrastructure into AI channels. Stripe is providing an agent-compatible payment layer. PayPal is developing its own agentic commerce infrastructure. Amazon is restricting third-party agent access to its marketplace.

The result is an emerging contest over the next digital distribution layer. For businesses, the strategic question may soon be less about whether they have an online storefront and more about whether their products, payments, and customer experience are accessible to the AI agents consumers use to navigate the internet.

Meta’s Muse AI Agent Tests Who Controls Digital Distribution

Agentic AIAgentic CommerceAI AgentAI agentsAI automationAI commerceAI DistributionAI in CommerceAI in paymentsAI Payment AgentsAI paymentsAI shoppingAI Shopping AgentAmazonArtificial Intelligencedigital commerceDigital Distributiondigital paymentsE-commercefinancial technologyfintech AIgenerative AIMetaMeta AIMeta AI AgentMeta MuseMeta Muse AIMuse AI Agentonline shoppingPayment infrastructurePayPalShopifyStripeTheFinRate